March 7, 2016
We’re only a day away from Primary Day in Michigan. It’s the best look we’re going to get at how things might set up on March 15. It’s not an exact match for any of the states, but when we see the result it will help us interpret places like Ohio, Illinois and Missouri.
When voters hit the polls for Super Tuesday, Donald Trump and Hillary Clinton were clearly in the driver’s seat here. Both led in every poll, both by double digit margins. Neither has blown a lead of this size in the final week of a contest.
Has anything happened to put either in peril? Failing at that, has another candidate made enough progress to indicate they are surging forward for March 15? We haven’t seen much of anything from those states since before Super Tuesday. Our clues are here.We have a combination of pollsters who have surveyed Michigan multiple times, and those who have weighed in more recently. It’s the best of both worlds. Enough recent versions to see if there’s a consensus among pollsters, enough history to see if the candidates are moving.
Let’s start with the Republicans:
Fox2 Detroit/Mitchell Research is the closest thing we have to a tracking poll. They just published a survey taken over the past four days. This is the post-debate poll. In addition, they took an immediately post-Super Tuesday survey and an on Super Tuesday version.
They are Trump-friendly. Not absurdly so. They have the highest reading for The Donald at any step along the way, but there’s usually someone else who winds up with very similar results. Those affirmations are from a range of different pollsters.
Here’s how he’s done with them over the past several weeks:
If you’re looking for a Trump is slipping argument, it’s not here. In late January, he was over 50% in a single poll, before some of the other candidates began getting traction in the state. After Iowa, he dropped to the low 40s, where he’s remained.
Before moving to adjust against other pollsters, let’s see how the other candidates have done over the same time. Oldest survey on the left, newest on the right.
Cruz: 15%, 16%, 11%, 16%, 14%, 19%, 19%
Mitchell doesn’t see Cruz as a strong contender to win, but he’s moving in the right direction after a dip in February. His final post-debate number is his best. Other pollsters have him higher, and he consistently out-performs on election day, so this is closer than it seems.
Kasich: N/A, N/A, 11%, 11%, 12%, 14%, 20%
This is an unambiguous trend. He was skipped in the first two surveys. Mitchell only asked voters about the top 6 candidates in national polls. Kasich wasn’t one of them. Then he sat in a non-competitive range for a couple weeks. There’s a definite post-debate bounce.
Rubio: 12%, 20%, 10%, 19%, 19%, 15%, 9%
Welcome to the Marco Rubio Roller Coaster. He started at 12% before Iowa, moved to 20% after he did well there. Then he fell on his ass in New Hampshire and plummeted to 10%. A strong recovery in South Carolina and he’s back to 19%. One dip after Super Tuesday, another post-debate.
It appears clear Cruz and Kasich have each taken voters from Marco. The more conservative have shifted to Ted, the more moderate/establishment-friendly to the governor. Another 10% of voters are undecided. When Rubio is doing well, many go to him, when he isn’t, they don’t.
Before taking other readings into account, it appears Trump is safe, Rubio is screwed, and undecided voters will determine whether Cruz or Kasich winds up second and how close to Trump they get.
Mitchell does robo-polling. They auto dial potential respondents and take them through an automated process. CBS/YouGov is an Internet-based survey. Both approaches are often criticized, but with small adjustments, they’re proving as predictive as any of the others.
So far, CBS/YouGov is Bernie-friendly on the Democratic side, lightly Trump-friendly on the GOP half, and somewhat pro-Cruz of late. They have Ted at 24%, his highest reading. Some of the poll was taken pre-debate, some after.
You can adjust down a couple points for their Ted-bias, but then would need to give him a couple extra points for the progress he’s made over the couple days since the survey ended. He also does well getting his voters out. Monmouth has him at 23% from 3/3-3/6.
Every poll except ARG (we’ll get to them in a minute) has Cruz between 19% and 24%. His move began at least a week ago though. Post-debate surveys do not have him noticeably higher than those taken immediately before.
There is a group of voters who choose between Rubio and Cruz. It appears Ted took most of the waverers several days ago and is retaining them. He also claimed any Ben Carson voters available to him between Super Tuesday and the debate.
After Cruz closed so strongly in the March 5 states (Maine, Kansas, Kentucky, Louisiana), we wondered if the same could happen in Michigan. It isn’t. There are three components to a strong Cruz showing.
First is taking voters from Rubio. Marco wound up with disappointing final numbers in each of the four states. Second is getting his vote out. Third is taking voters from Trump. When Cruz went from losing Louisiana absentee voters by 25 to tying Trump with election day voters, he did all three.
Polls show he’s taken care of the Rubio part in Michigan, but there isn’t much more to grab from him. Trump isn’t losing any support. Those who consider Ted against The Donald are keeping with their original thought. Even with a strong ground game, Cruz is missing one of the three components of closing a big gap.
Kasich began taking Rubio voters after the debate. Prior, he was under 20% in every poll, trailing Marco in most of them. That’s a crucial line for Kasich. Over it, you can add a dose of home field advantage and project him to a win in Ohio. Below, not so much.
Pre-Debate: 14%, 13%, 18%
Post-Debate: 33%, 20%, 21%
The 33% is from ARG, which is regularly more optimistic about Kasich than his own internal polls. They’re always high on him. It’s still twice as much support as they found for him two weeks before. He’s making progress regardless of what level he’s at.
He looks like a good bet to wind up on the right side of the 20% barrier. While the momentum is in his favor, he almost never exceeds his polling average in actual voting. You can average in ARG as a data point, but expecting him to leap another several points forward in the final 48 hours is probably a mistake.
We’ll attack this in a separate piece, but Cruz and Kasich sort of need each other to get closer to Trump. It appears Cruz overlaps with Trump and Rubio. Kasich pulls from Rubio and Trump. Few voters (for now) are deciding between Cruz and Kasich. That would only happen with Marco gone.
That means any time Cruz takes a voter from Trump, it pulls The Donald closer to where Kasich could catch him in a Midwestern state. Same goes for when Kasich steals a voter who thinks Trump is too coarse, but Cruz too conservative.
It doesn’t look like either are doing well enough with Trump voters to pull him down to the 30-32% level where someone could beat him. Aside from ARG (31%), no pollster has him below 36%.
Michigan is an open primary. Trump does best in these. A closed caucus is the toughest, with open caucus (which don’t really exist) and closed primary fitting in between.
He also does better when polls have consistently shown him with large margins over several months. Apologies for making you sift through this many words to get here, but despite any progress made by Cruz or Kasich, I don’t see the math indicating a Trump defeat.
Time for the Democrats.
A two person race is always easier to track. Nothing like having fewer variables to sift through. Mitchell does not like Bernie. Most states have shown two distinct polling tracks. There’s usually a group far more favorable to Hillary, another more optimistic for Bernie.
In states where Sanders has a chance, the optimistic polls are usually more accurate. He tends to wind up somewhere between the mid-point and the rosy scenario. Down South, the reverse is true.
It appears these pollsters are modelling different electorates. Hillary-friendly territories are more like one, Bernie-friendly like the other. Michigan is interesting because it combines characteristics of both.
Tomorrow we find out if this means we can just take the mid-point of the polling outcomes. Until then, there are two clear messages:
One: Hillary is leading. The closest poll taken in 2016 has him trailing by 10.
Two: Bernie is sliding. Her margins with common pollsters are higher than 7 to 10 days ago.
CBS/YouGov tends to favor Sanders. They had Hillary up 11 from 3/2-3/4.
ARG is more neutral, and had Hillary up 13 from 2/19-2/20, but up 24 from 3/4-3/5.
I’d expect Mitchell to overstate Clinton’s margin, but they’ve gone in the following direction from 1/25 to the present:
+27, +29, +33, +34, +28, +37
She currently has her largest edge of 2016. If Bernie isn’t ahead anywhere and isn’t going in the right direction anywhere, he’s going to lose by a noticeable margin.
Monmouth released their first Michigan survey, taken from 3/3 to 3/6 today. Hillary is ahead 55/42, a relatively modest gap. If they have a bias, it’s apparently in the direction of closer results.
They were very mildly pro-Hillary in Iowa, very pro-Hillary in New Hampshire, very pro-Bernie in South Carolina. In the latter two states, the final margin was about double what they expected.
If you throw all of this together, and treat this as a state with some traits that favor each, the evidence is pointing to a plus/minus tomorrow of 20 points.
We’ll issue our official predictions in the morning, but with plenty of data to look at here, you can definitely see where we’re leaning.